Introduction

Nitrogen (N) removal in sewage is one of the main goals of biological sewage treatment. High N emissions can cause accelerated eutrophication in a receiving water body (De Sotto et al. 2018). The sequencing batch reactor (SBR) activated sludge process is commonly used in rural sewage treatment plants, offering advantages in terms of investment, fluctuation of sewage characteristic and volume, energy consumption, as well as good removal of organic matter and phosphate (Maslon et al. 2019). Nevertheless, the effluent on total N (TN) would not meet the first class requirement of the National Municipal Wastewater Discharge Standards of China (NMWDS) (TN < 10 mg L−1) (Yuan et al. 2016; Chen et al. 2016). Reliable TN removal requires adequate and stable denitrifier populations, suitable carbon sources, and adequate duration for anoxic denitrification (Onnis-Hayden et al. 2011). Thus, process modifications were applied that were related to the reaction period of SBR, the way in which the sewage and oxygen interact, and the inclusion of a suspended biofilm carrier (Mihelcic and Zimmerman 2014; Zhang et al. 2019a, b, c, d; Tombola et al. 2019; Tao and Hamouda 2019). Moving bed biofilm reactors (MBBRs) have been shown to be successful for the enhancement of nitrification and denitrification in a WWTP upgrade project (Onnis-Hayden et al. 2011; Makowska and Maciejewska 2016; Sytek-Szmeichel et al. 2016). To obtain sufficient mixing and interaction between sewage and the suspended biofilm carrier in MBBRs, high intensity aeration is necessary, causing drastic collisions among the carrier and the inner wall of the MBBR. In a full-scale MBBR, it was found that (1) broken carriers and its debris caused clogging of the subsequent clarifier and (2) abrasion of the MBBR inner wall (see images presented in Supplementary Information Fig. S1). Prior to maintenance of the aeration system (ca. every two years), all of the carriers need to be cleared from the system, resulting in great losses in carrier quantity. All these problems mentioned above need to be solved for efficient full-scale application of MBBRs.

To shun problems on carrier loss and operation, the semi-fixed biofilm carrier was introduced in this study, as well as decoupling the growth rate of nitrifying populations and the suspended mixed liquor phase solids retention time. Meanwhile, the sludge elevation mixing and the change on the mode of influent and effluent were also conducted, obtaining reliable TN removal for both low- and high-strength sewage treatment. The performance of our novel SBR would be attributed to the joint effect of microorganisms, including ammonia oxidizing bacteria (AOB), nitrite oxidizing bacteria (NOB), denitrifying bacteria (DNB) (Rud et al. 2017) and other heterotrophic bacteria. Although analysis of the classical bacterial 16S rRNA gene-based marker can reveal community composition, this type of analysis will not provide direct functional analysis of TN removal in this novel SBR. Recently, the microbial community and its hydrolase functional profile in the mainstream upflow nitritation-anammox system with hybrid anaerobic pretreatment (Li et al. 2017b), sewage treatment systems (Cui et al. 2019; Zhang et al. 2019a, b, c, d; Zhao et al. 2020), sewage sludge, and manure composting processes (Jiang et al. 2019; Zhong et al. 2020) were analyzed using Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt)—a widely used algorithm which optimizes genome prediction that enables improved accuracy on functional profiling of complex phenotypes and integration to customary databases (Douglas et al. 2019). Therefore, to understand the microbial community composition from suspended aggregates and semi-fixed biofilm carrier, the change in community structure related to variable organic load, and their role in the N removal pathway, PICRUSt2 (the latest version) was applied in this work.

The aim of this study is to (1) investigate the bacterial community from suspended aggregates and semi-fixed biofilm carrier during low- and high-strength sewage treatment; (2) explore the biomarkers of N removal in the bacterial community; (3) screen the possible functional shifts of the bacterial community driven by changes in sewage load strength; and (4) understand the interactions among microbial species. By means of high-throughput sequencing technology, principal component analysis (PCA), and PICRUSt2, our result will provide insights into the bacterial community from suspended aggregates and semi-fixed biofilm carrier in terms of phylogenetic structure and functional profiles during low- and high-strength sewage treatment in this novel SBR.

Materials and methods

SBR pilot plant operation and sewage quality monitoring

The pilot-scale SBR integrated with semi-fixed biofilm carrier (Fig. 1) was set up in December 2017 at a sewage treatment plant in Guangzhou (China), and was operated for over 300 days with a capacity of 90 m3 day−1. As depicted in Fig. 1, four improvements, including semi-fixed biofilm carrier, gravity influent, sludge elevation mixing, and gravity effluent, were introduced. The semi-fixed biofilm carrier (Supplementary Information Fig. S2d) is composed of plastic film clip and vinylon fibers (Jiangsu Dadu plastic CO. LTD, Zhejiang, China), which bundled in steel frame and fixed in the lower half of the reactor (Fig. 1). The other major equipment includes an aeration system, mixer, diffuser, and other components. The seeding sludge was introduced from sewage treatment plant, while SRT was set approximately 25 days. Operation of this reactor was controlled automatically by a system designed for pilot-scale operation and was set on-site. This automatic system controls the electric pumps and air valves that were used for the flow of water and sludge, respectively into and out of the reactor. The total cycle was 8 h, with the main process following the cycle of influent (~ 10 min), first stage return of activated sludge elevation mixing under anaerobic conditions (~ 50 min, DO < 0.2 mg L−1), aeration (~ 120 min with air flow ratio of 1000–3000 L h−1, DO 1.5–3.0 mg L−1), second stage return of activated sludge elevation mixing under anoxic conditions (~ 210 min, DO < 0.5 mg L−1), dosing for the removal of phosphorus (~ 10 min), sedimentation (~ 70 min), and effluent (~ 10 min). The feed water was imported from primary sedimentation effluent (defined as low-strength sewage, 48–224 mg L−1 for chemical oxygen demand (COD) and 10–30 mg L−1 for TN, respectively), stored in head tank and then flowed into the reactor in the form of gravity flow via wide tube. To test the performance on high-strength sewage (ca. 64–482 mg L−1 for COD and 16–57 mg L−1 for TN, respectively), condensed wastewater from a septic tank was introduced and mixed with the primary sedimentation effluent between August and October 2018. The main characteristics of the influent sewage applied in this work are described in Supplementary Information Table S1.

Fig. 1
figure 1

Diagram of the novel sequencing batch reactor integrated with semi-fixed biofilm carrier reactor

Full size image

The influent sewage and effluent were collected daily in triplicate. And the COD, ammonia, TN, and total phosphates (TP) tests were conducted using commercial kits from Hach® following the manufacturer's instructions. Briefly, filtered samples were diluted (for TP, TN, and ammonia tests) individually in tubes. COD and TP samples were digested using a heat block at 150 °C for 2 h and 30 min, respectively, while TN was digested for 30 min at 105 °C. The digested samples were cooled down before analysis. Concentrations were identified using the default programs of a DR/2500 spectrophotometer (Hach, Loveland, CO, USA).

Sample preparation

Suspended aggregates (SU) and biofilms (B), named as L_SU, L_B, H_SU, and H_B from low- and high-strength sewage loads in May and October, respectively, were collected independently at the end of the anoxic period (Fig. 1); samples were collected in triplicate using sterile tubes and pooled. To isolate the biofilm biomass from detachable carriers, the carrier was snipped into fragments and then mixed with lysozyme at 35 °C for 20 min. Afterwards, the biofilm was separated from the carrier using a Vortex oscillator (GL-88B, Jiangsu Haimen Kirin Medical Instrument factory). After removing the carrier, the remaining suspension was used as the biofilm sample. Samples L_SU, L_B, H_SU, and H_B were centrifuged at 15,000 × g for 15 min and stored – 20 °C until DNA extraction.

DNA extraction and bacterial community analysis

PICRUSt2 was used to predict the metabolic dynamics of the communities (Li et al. 2017b). The protocols for DNA extraction, high-throughput 16S rRNA gene pyrosequencing, and biodiversity analysis are presented in Supplementary Information Text S2.

Data analysis

The COD, ammonia, TN, and TP values from influent and effluent were analyzed using Origin 10.0 software (Origin Lab, Northampton, USA). A high-throughput sequencing technology, PCA and PICRUSt2, was used to investigate the bacterial community.

Results and discussion

Nutrient removal efficiency

After startup (January to March 2018), this novel SBR exhibited reliable treatment performance for the removal of TN (Fig. 2), COD, and ammonia (Supplementary Information Text. S3) at COD/TN influent ratio < 7.5. The effluent met the first class requirement of the NMWDS. Even the sewage strength load soar, minor differences from sewage parameters between low- and high-strength sewage load at aerobic, anoxic, and effluent stage (Supplementary Information Fig. S9) would show that this novel SBR is available to load impact. Moreover, the reactor presented excellent TN removal during high-strength sewage load, with average values of 11.66 mg L−1 for effluent, comparing to that of 7.33 mg L−1 during low-strength sewage load. It was assumed that semi-fixed biofilm carrier and the sludge elevation mixing offer abundant C and biomass to cope with the augment of sewage strength for denitrification. On the other hand, the hydraulic retention time for denitrification (~ 330 min), which occupied 67.3% of the total operation duration, would also favor TN removal. Herein, gravity influent was applied to replace common pressure flow by pump. The total duration for influent in this novel SBR was merely 10 min (Fig. 1), which is far less than that of pressure flow (30 min ~ 60 min). Similarly, gravity effluent via wide tube was another improvement to save reaction time. The improvements on influent and effluent mode would free up ca. 40 min ~ 100 min for anoxic denitrification. However, the poor TP removal from July to September (some plots > 0.5 mg L−1 in Supplementary Information Fig. S8) was because of the absence of dosing poly aluminum chloride.

Fig. 2
figure 2

Total nitrogen (TN) concentrations in a influent (and the values of May and October in the insert box chart, respectively represented for low- and high-strength sewage load); b effluent. This novel SBR had been operated with three stages, including startup (from January to April), low TN influent (from April to August), and high TN influent (from August to November)

Full size image

Microbial communities in suspended aggregates and biofilms

The microbial communities in suspended aggregates and in attached biofilm on carrier within this novel SBR were investigated. After quality filtering, the diversity of OTUs, calculated by estimating the number of total OTUs at the 97% similarity level by the Shannon diversity index, was about 5.98, 5.95, 5.73, and 5.76 in L_SU, L_B, H_SU, and H_B, respectively (Supplementary Information Table S2), indicating that community diversity was higher in low-strength sewage than in high-strength sewage. Analysis of the shared OTUs among samples showed that there were 852 OTUs observed in both samples while more OTUs were present in low-strength sewage than in high-strength sewage (Supplementary Information Fig. S10). This could be explained by the leap of some strains, which is fond of high-strength sewage and become dominant. Besides, the different bacterial composition was also analyzed by PCA based on the abundance of OTUs. The first axis (F1) and the second axis (F2) accounted for 81.4% and 12.1% of the variability, respectively (Supplementary Information Fig. S11). PCA revealed the difference in community composition between suspended aggregates and attached biofilm. H_SU and H_B communities were more distant from each other than those of L_SU and L_B communities. These results suggest that the bacterial compositions of H_SU and H_B were more dissimilar thus facilitating cooperation during treatment of high-strength sewage.

Taking a closer look at the microbial community constituting the suspended and attached growth is necessary to understand the association between these two forms of growth and the influence of microorganisms in suspended aggregates on the formation of attached biofilm on carriers, and to eventually elaborate the evolution of microorganisms during the change in sewage strength. A total of nine phyla were identified in the samples (Fig. 3). The abundance of the OTUs in different phyla was listed in order as follows: Proteobacteria, Bacteroidetes, Acidobacteria, Planctomycetes, and Chloroflexi. Phyla Acidobacteria and Chloroflexi are known to contribute to carbon and N cycles (Eichorst et al. 2018; Nguyen, et al. 2019). Meanwhile, previous studies (Chen et al. 2018; Lage et al. 2019) have linked the presence of Proteobacteria, Bacteroidetes, and Planctomycetes to nitrite denitrification.

Fig. 3
figure 3

Relative abundance of classes (and phyla) in suspended aggregates during low-strength sewage load stage (L_SU), biofilms during low-strength sewage load stage (L_B), suspended aggregates during high-strength sewage load stage (H_SU), and biofilms during high-strength sewage load stage (H_B)

Full size image

To get a better understanding of the N removal pathway, the group diversity was measured to discriminate between communities in suspended aggregates and attached biofilm during low- and high-strength sewage load treatment. Figure 4 is a hierarchical clustering analysis of communities in L_SU, L_B, H_SU, and H_B. A positive z-score (red) denotes that the value in that sample is above the mean of the group while negative z-scores (blue) are for those values that are below the mean. The predominant bacterial community consisted of 50 genera (Fig. 4a) mainly from phylum Proteobacteria (20 out of 50) followed by Bacteroidetes (9 out of 50) and Acidobacteria (5 out of 50). The group diversities observed during low- and high-strength sewage treatment were diverse and were present at different abundances as well as in suspended aggregates and attached biofilm for the same sewage load (Fig. 4a). The different communities could exert different contributions to the removal/biotransformation of compounds containing N, P, and pollutants like recalcitrant hydrocarbons because of their metabolic and biochemical capabilities. In detail, L_SU was mainly characterized by Nitrospira, Rhodocyclaceae, and Anaerolineaceae. The genus Nitrospira are prevalent NOB in most sewage treatment systems (Shu et al. 2015), while Rhodocyclaceae and Anaerolineaceae are capable of autotrophic denitrification (Guo et al. 2016) and carbohydrate decomposition under anaerobic conditions (Narihiro et al. 2012), respectively. When compared with L_SU, the predominant strains in L_B were mainly those capable of ammonium oxidation (Pirellulaceae (Miao et al. 2019) and Ellin6067 (Xia et al. 2005)) and organic contaminants degradation (Bacteroidetes (Chen et al. 2007) and Sphingomonadaceae (Thelusmond et al. 2016)). Besides Rhodocyclaceae, some denitrifying strains (including Haliangium (McIlroy et al. 2014) and Dechloromonas (Zhang et al. 2019a, b, c, d)) were present at higher abundance in L_SU than in L_B. The processes of denitrification (represented by DNB) and nitrite oxidation (represented by NOB) were more prevalent in L_SU than in L_B, whereas the opposite was the case for the process of ammonia oxidation (represented by AOB; Fig. 4b). From these results, it was concluded that the microbial community of L_SU played a more important role in denitrification than that of L_B during low-strength sewage treatment. With the increase in sewage strength, some strains associated with organic contaminants degradation (including Fimbriimonadaceae (Feng et al. 2016) and Novosphingobium (Liu et al. 2005)) became dominant in H_SU (Fig. 4a), while strains with unknown function became dominant in H_B, such as Chitinophagales, AKYH767, SBR1031, RBG-13–54-9_g_2. The presence of SM1A02 strains has been widely detected in anammox systems (Tian et al. 2017; Zhang et al. 2018; Meng et al. 2019; Gao et al. 2020), indicating that anammox process potential was a crucial link in the N metabolism pathway. Anammox is a biologically mediated process in which ammonium is oxidized to N2 gas under anaerobic conditions with nitrite serving as the electron acceptor (Mulder et al. 1995; Strous et al. 2006). Li et al. (2020) investigated two anammox granular sludge reactors operating under different N loading rates and found that high TN removal was observed under high N loading. Herein, the average TN removal during high-strength sewage treatment was as high as 73.47%, whereas lower average TN removal rate (63.41%) was recorded during low-strength sewage treatment (calculated from Fig. 2). Besides, ampler nitrite at anaerobic stage during high-strength load (Supplementary Information Fig. S9) would provide favorable substrate for anammox.

Fig. 4
figure 4

Hierarchical cluster analysis of bacterial communities. The relative abundance of a 50 most abundant operational taxonomic units (OTUs) and b functional community of anaerobic ammonia oxidation (ANAMMOX), ammonia-oxidizing bacteria (AOB), denitrifying bacteria (DNB) and nitrite-oxidizing bacteria (NOB)

Full size image

Mechanism of N removal

To illustrate the N metabolism during low- and high-strength sewage treatment, the PICRUSt2 algorithm and KEGG database (Douglas et al. 2019) were applied to categorize the functional genes distribution and the enzyme families’ abundance, respectively. Based on the reference pathway (KO) of N metabolism (map00910) (Kanehisa et al. 2016), the N metabolism pathways were mapped as Supplementary Information Fig. S12, and then subdivided as the core (Supplementary Information Fig. S13) and the subordinate (Supplementary Information Fig. S14) pathways. Referring to Kanehisa et al. (2016), four reduction pathways (M00175 for N fixation, M00531 for assimilatory nitrate reduction, M00530 for dissimilatory nitrate reduction, and M00529 for denitrification) and one oxidation pathway (M00528 for nitrification) were involved in the core N metabolism pathway (Supplementary Information Fig. S15). The modules of M00528 and M00531 showed higher abundance during high-strength than low-strength sewage treatment, while M00175, M00530, and M00529 were the opposite. The increase in sewage loading reduced the contribution of autotrophic nitrate reducing bacteria and weakened their dominant role, even though the TN removal rate was high during high-strength sewage treatment (see “Microbial communities in suspended aggregates and biofilms” section). These results were similar to those of Qiu et al. (2020), which demonstrated that the relative abundance of denitrifiers decreased with increasing organic load during long-term operation in a sulfur-based denitrification reactor. In our study, the increase in sewage strength boosted the total microbial abundance, provided favorable habitat for microbial communities that were not related to autotrophic denitrification, and then chopped the total relative abundance of denitrifier and functional genes (Miao et al. 2019; Gu et al. 2019; Sun et al. 2020). These conclusions were supported by the high abundance of M00531 (assimilatory nitrate reduction pathway) and the low abundance of M00804 (complete nitrification pathway) during high-strength sewage treatment (Supplementary Information Fig. S15). The lower M00804 during high-strength sewage load would result in an increase in nitrite formation, which is one of the substances required for anammox (Liu et al. 2020). This data would support the potential for anammox during high-strength (see “Microbial communities in suspended aggregates and biofilms” section) rather than low-strength sewage load. However, because of the lack of key enzyme genes in the KEGG database, the predictive functional profiling for anammox could not be concluded in this study. In brief, the difference in N metabolism between low- and high-strength sewage loads was explained by the prevalence of nitrification–denitrification for TN removal during low-strength sewage load, whereas assimilation was prevalent during high-strength sewage load.

In combination with the key enzyme coding (Li et al. 2019; Yang et al. 2020) presented in this study, the main N metabolism processes were shown in Fig. 5. Several key enzymes were involved in the nitrification–denitrification pathway, such as ammonia monooxygenase (EC:1.14.99.39) and hydroxylamine dehydrogenase (EC:1.7.2.6) responsible for nitrosation (Fig. 5), nitrate reductase EC:1.7.7.2, EC:1.7.5.1, EC:1.7.6.1 and EC:1.7.99, involved in the redox between nitrate and nitrite, nitrite reductase (EC:1.7.2.1) related to the reduction of nitrite to nitric oxide, nitric oxide reductase (EC:1.7.2.5) associated with the reduction of nitric oxide to nitrous oxide, and nitrous oxide reductase (EC:1.7.2.4) responsible for the reduction of nitrous oxide to N2. With the increase in sewage strength, EC:1.7.2.1 and EC:1.7.2.5 showed an obvious decrease; meanwhile, glutamate dehydrogenase (EC:1.4.1.3 and EC:1.4.1.4) and glutamate-ammonia ligase (EC:6.3.1.2) were both increased. These indicated that during high-strength sewage load, first, there was a whittle for denitrification pathway; second, TN removal was mainly through the formation of L-glutamine via assimilation. These results are similar to those of Zhang et al. (2019a, b, c, d) and Wang et al. (2020), who documented that organic supplementation enhance the assimilation efficiency of ammonia-N and boosted the conversion of ammonia-N to protein. In contrast, because the enzymes involved in nitrosation (EC:1.14.99.39 and EC:1.7.2.6) had far lower abundance than other enzymes (Fig. 5), it seems that the reaction dynamic was constrained. These results were not consistent with the long, reliable, and compliant standard for TN discharge during low- and high-strength sewage loads, respectively. This might be because the samples were collected at the end of the anoxic period (Fig. 1), when nitrosation microorganisms struggled to survive and showed weak endurance (Guo et al. 2009; Cui et al. 2020).

Fig. 5
figure 5

Abundances of key enzyme classifications (identified by their EC number) for each treatment in the nitrogen metabolism pathway

Full size image

To illustrate the role of suspended aggregates and biofilm carrier involved in TN removal, the proportions of the key functional genes in suspension and biofilm samples related to the main N metabolism pathways were itemized in Fig. 6 and Supplementary Information Fig. S16. The nitrate reductase genes in L_SU (i.e., nirS, norB, nosZ, narG, narH, narI, napA, nirB, and nirD) were present at high abundance, indicating the dominant role of L_SU in TN removal, while these genes were not prominent in L_B. With increasing sewage strength load, the abundance of nitrate reductase genes in H_SU decreased. Meanwhile, the genes (including nasB, nasA, nirA, and glnA) related to N assimilation in H_B became dominant. The glnA gene is the key gene encoding the enzyme for glutamate synthesis during N uptake via microbial growth and protein synthesis (Matassa et al. 2016; Li et al. 2017a, b). The change in the key genes associated with TN removal was related to the condition of heterogeneous strains, which were less competitive than autotrophic bacteria during low-strength sewage load but tended to increase in abundance with increased nutrient content to impel ammonia assimilation and assimilatory nitrate reduction in this study (Qiu et al. 2020). The biofilm carrier provided the habitat for heterogeneous strains and then enhanced the assimilatory nitrate reduction pathway for protein synthesis during high-strength sewage load. Moreover, with the presence of Limnodrilus only during high-strength sewage load (Supplementary Information Fig. S2), predation would be the subsequent pathway after the assimilatory nitrate reduction in this work. But it needs more data to support this postulation.

Fig. 6
figure 6

Heatmap of proportions of the key functional genes in suspension and biofilm samples related to the main nitrogen metabolism pathways

Full size image

Conclusion

This study investigated the microbial community and the N removal pathway in the suspended aggregates and biofilm carrier of a novel SBR configuration, during treatment of low- and high-strength sewage loads. The value of weighted NSTI for L_SU, L_B, H_SU, and H_B were less than 0.35, indicating PICRUSt2 in well predicting a high correlation to the metagenome sequencing data on our microbial samples. The reliable TN removal attributed to the introduction of semi-fixed biofilm carrier, the change on the mode of influent and effluent to free up extra reaction time for denitrification, and the sludge elevation mixing during anaerobic and anoxia stage. The suspended aggregates and biofilm carrier played different roles in N pathway. During low-strength sewage loads, the suspended aggregates were responsible for denitrification, while the biofilm took charge of ammonium oxidation. With the augment on sewage strength loads, the role of nitrate reductase from suspended aggregates faded, and anammox and N assimilation by biofilm became dominant. The TN removal during low-strength sewage loads was mainly by nitrification–denitrification, while the formation of l-glutamine via assimilation was prevalent during high-strength sewage loads. These analyses classified the microorganisms that thrive in the different compartments of this novel SBR and highlighted possible relationships between microbial community structure during TN removal.